Human may violate known rules(or even laws) temporarily if all known legal solution can't get itself out of the problem. But AI nowadays don't do (or the design

Human may violate known rules(or even laws) temporarily if all known legal solution can't get itself out of the problem. But AI nowadays don't do (or the designer of AI aren't allowed to? Is that even legal to do?). And besides that. If such decision actually cause injury eventually, who should response for that? The AI designer? The car company? The AI itself? The car owner? We still don't have a definitive answer for that yet.